Final Project

Set up

Loading Packages

Cleaning Data

We want to investigate the the searches that is from the US, is using Expendia.com, is from not mobile site, and is not from a package research for a domestic hotel that is branded, has star rating above 4. The check-in and check-out dates should be within the summer vacation period (i.e., from June to August) in 2015.

data <- read_delim("data.txt", delim = "\t", escape_double = FALSE, trim_ws = TRUE) %>%
  drop_na() %>%
  filter(user_location_country == "UNITED STATES OF AMERICA" & site_name == "EXPEDIA.COM" & is_mobile == 0 & is_package == 0 & hotel_country == "UNITED STATES OF AMERICA") %>%
  filter(prop_is_branded == 1) %>%
  filter(prop_starrating == 4 | prop_starrating == 5) %>%
  mutate(srch_ci = ymd(srch_ci, tz = "America/New_York"), 
         srch_co = ymd(srch_co, tz = "America/New_York")) %>%
  filter(srch_ci %within% interval(ymd_hms("2015-06-01 00:00:00 EDT"), ymd_hms("2015-08-31 23:59:59 EDT")) & srch_co %within% interval(ymd_hms("2015-06-01 00:00:00 EDT"), ymd_hms("2015-08-31 23:59:59 EDT")))

write.csv(data, "new_data.csv")

Importing Data Set

# When re-opening this file, only run chunks after this. 
data <- read.csv("new_data.csv")
dest <- read_delim("dest.txt", delim = "\t", escape_double = FALSE, trim_ws = TRUE)
state_pops <- read.csv("us-state-populations.csv")
party <- read.csv("party.csv")

Creating Sample Data Set

sample_data <- data[sample(1:319911, 1000, replace = FALSE), ]

Data Processing and Presentation

Data Inclusion Criteria:

• US users
• using Expendia.com
• not from mobile site
• not from a package research
• searching for a domestic hotel
• hotel is branded
• hotel has a star rating above 4
• check-in date from June 1st, 2015 to Augest 31st, 2015
• check-out date from June 1st, 2015 to Augest 31st, 2015

Shiny App on the distribution of popular_food_

food <- read.csv("new_data.csv") %>% 
  filter(is_booking == 1) %>%
  inner_join(dest, by = "srch_destination_id") %>%
  pivot_longer(col = starts_with("popular_food"), names_to = "popular", values_to = "probability") %>%
  select(popular, probability)

names <- food %>%
  distinct(popular)

ui <- fluidPage(
    titlePanel("Compare Three Popular Scores"),
    selectInput(inputId = "popular1", 
                label = "Popular Score 1", 
                choices = names), 
    selectInput(inputId = "popular2", 
                label = "Popular Score 2", 
                choices = names), 
    selectInput(inputId = "popular3", 
                label = "Popular Score 3", 
                choices = names), 
    mainPanel(plotlyOutput(outputId = "boxplot"))
)

server <- function(input, output) {
  output$boxplot <- renderPlotly({
    food %>%
      filter(popular == input$popular1 | popular == input$popular2 | popular == input$popular3) %>%
        plot_ly(x = ~popular, y = ~probability, type = "box", color = ~popular, boxpoints = FALSE)
    })
}

# Run the application 
shinyApp(ui = ui, server = server)

Shiny applications not supported in static R Markdown documents

Map showing the per mil of number of bookings per population in different states of US.

map_data <- map_data("state") %>%
  mutate(region = toupper(substr(region, 1, 2))) %>%
  distinct(region, .keep_all = TRUE)

map <- data %>%
  group_by(user_location_region) %>%
  mutate(n = sum(is_booking)) %>%
  select(user_location_region, n) %>%
  distinct(user_location_region, .keep_all = TRUE) %>%
  left_join(state_pops, by=c( "user_location_region" = "code")) %>%
  mutate(rate = n / pop_2014 * 1000) %>%
  left_join(party, by = c("user_location_region" = "state_code")) %>%
  select(user_location_region, rate, party)

plot_geo(map, locationmode = "USA-states") %>%
  add_trace(locations = ~user_location_region, type = "choropleth", z = ~rate, text = ~party, colorscale = "Reds")

A heat map comparing distance_band and hist_price_band with regard to the number of bookings.

dat <- data %>%
  select(distance_band, hist_price_band, is_booking) %>%
  group_by(distance_band, hist_price_band) %>%
  count(is_booking) %>%
  filter(is_booking == 1)

dat$distance_band <- factor(dat$distance_band, levels = c("VC", "C", "M", "F", "VF"))
dat$hist_price_band <- factor(dat$hist_price_band, levels = c("VL", "L", "M", "H", "VH"))

dat %>%
  plot_ly(x = ~factor(distance_band), y = ~factor(hist_price_band), z = ~n, colorscale = "Reds") %>%
  add_heatmap()